Hello!
The topic you're trying to explain here is called human detection in frames or scenes. There have been many research on this, and clearly neural network is currently the best at doing such things, however due to complexity in training and testing and storing such large networks and millions of parameters, computer vision also has image processing based methods for doing the same. The two moat popular ones are Viola Jones and HOG features. The both of them are perhaps the most cited papers in this field after published in 2005.
So, the same can be implemented in your case too for achieving similar results.
About me:
I work as a research student in this field and take up projects that meet my ine of work from time to time. We can use MATLAB for development purpose and if required can later convert it to c++ with opencv for embedded systems work (that is however out of scope of this project)
Deliverables:
1. Code
2. Datasets
3. Results & Assistance
Challenges:
There are few challenges to address (conceptually), can explain during chat
If you're interested, kindly message me whenever you're free. If not online, shall reply you ASAP.
Thank you! Have a nice day!